An Electron-Diffraction Investigation of Nitrogen Trifluoride1 - Journal

An Electron-Diffraction Investigation of Nitrogen Trifluoride1. Verner Schomaker, and ... Nitrogen Fluorides and their Organic Derivatives. Charles J...
0 downloads 0 Views 451KB Size
VERNEKSCHOMAKER AND CHIA-SI Lu

1182 [ CONTRIBUTION FROM

THE

VOl. 72

GATESAND CRELLIN LABORATORIES OF CHEMISTRY,CALIFORNIA INSTITUTEOF TECHNOLOGY. No. 13081

An Electron-Diffraction Investigation of Nitrogen Trifluoride' BY VERNERSCHOMAKER AND CHIA-SILu2

Introduction and Discussion.-Since its first preparationa in 1928 by Otto Ruff, nitrogen trifluoride has been of obvious interest to chemists because i t is one of the very few known substanoes containing the nitrogen-fluorine bond. Although Ruff and his associates continued the study of the preparation and reactions of nitrogen trifluoride and measured its heat of formation, density, melting point, boiling point and vapor pressures, little other work has been reported. Watson, Kane and Ramaswamy found the electric only about 0.2 X 10-l8 moment to be very e. s. u., and concluded from this value and from a comparison with the value for ammonia ( ~ N H=~ 1.46 X lo-'* e. s. u.), that the molecule is nearly, but not quite, planar. Bailey, Hale and Thompson5 reported values and assignments for three of the vibrational frequencies of the molecule from observations of the infrared spectrum, derived the value 4.10 X lo5 dynes/cm. for the stretching force-constant of the N-F bond (and from this 1.45 A. for the bond length according to Badger's rules) and concluded, perhaps with the help of a normal-coordinate treatment for a special potential function or a study of band contours, that the molecule is a very flat pyramid with bond angle greater than 110'. Neither of these structural conclusions is accurate, as we shall see, but the electric moments mentioned above are interesting and will be discussed briefly, in connection with the results of our work. According to our electron-diffraction investigation, the nitrogen trifluoride molecule, assumed to be a symmetrical pyramid, has a bond angle ( L F N F ) of 102.5 * 1.5', a bond distance (IN-F) of 1.37 rf; 0.02 B.,and a non-bond distance ( I F . . . F ) of 2.14 * 0.02 A. We have made a normal coordinate treatment for a molecular model with these parameters. On neglecting the interaction force constants involving adjacent N-F stretchings, adjacent F-N-F bendings, and non-adjacent stretching and bending, the following force constants are found to correspond to the observed frequencies :a N-F stretching, 3.724 X lo6 dynes/cm.; F-N-F bending, 1.434 X lo-" dyne-cm./radian2; stretching-bending interaction, (1) This article is based on work done under Contract OEMsr753 between the National Defense Research Committee and the California Institute of Technology. (2) Present address: National University of Amoy. Amoy, Fukien; China. (3) 0. Ruff, J. Fischer and F. Luft, Z . anorg. allgcm. Chcm., ill, 417 (1928). (4) H. E. Wntson, G. P. Kane and K. L. Ramaswamy, Proc. Roy. SOC.(London), Ai68, 137 (1938). (6) C. R.Bniley, J. B. Hale and J. W. Thompson, J . Chcm. Phys., I , a74 (may). (6) R. M.Badger, ibid., 8 , 710 (1936).

0.015 X lo-* dyne/radian. The unobserved fundamental frequency is calculated to be 400 ern.-'; Bailey, et aL16gave an estimate of 420 ern.-'. Since our values for the force constants are reasonable (the magnitude of the interaction constant may, to be sure, not have any significance) the conclusion regarding the molecular configuration of nitrogen fluoride arrived a t by Bailey, et u Z . , ~ on the basis of the infrared spectra is a t least not demanded by the frequencies of the observed bands. As stated above, the molecule is by no means a nearly flat pyramid, but instead has a less than tetrahedral bond angle comparable to those observed7 for other fifth group halides and for oxygen fluoride and appreciably less than the approximately tetrahedral value found in the amines, ammonia, the ethers and water. We believe that the surprisingly small value of the electric moment of nitrogen trifluoride may nevertheless be accounted for, as the result of an almost perfect cancellation of the expected moment due to the ionic character of the bonds by a moment of opposite sign, arising from such further electrical asymmetries of the molecule as would be present even with purely covalent bonds, which is mainly due in this case to an imperfect balancing of the electric moment contribution of the unshared electrons of the nitrogen and fluorine atoms. This lack of balance could result from a difference* in the hybrid character of the nitrogen and fluorine bond orbitals. I n the ammonia molecule, on the other hand, the resultant ionic moment is reversed (with presumably little change in magnitude), so that the two contributions are of the same sign, and a large total moment is observed. The suggested interpretation of the electric moments of ammonia and nitrogen trifluoride may be of significance also in connection with the bond angle values. It is notable that when a nitrogen or oxygen atom is singly bonded to less electronegative atoms such as carbon or hydrogen, the bond angle tends to be larger than when the nitrogen or oxygen atom is bonded to the more electronegative fluorine atom. We suggest that part or all of this effect may arise, in the way de(7) See L. Pauling, "The Nature of the Chemical Bond," 2nd edition, Cornel1 University Press, Ithaca, New York, 1940, p. 79. (8) If, for example, the nitrogen bond orbital were of the pure tetrahedral, spa, type, while the fluorine bond orbital was an unhybridized p orbital with zero moment, its contribution would be very large and unbalanced. The resultant of the moment of all the orbitals would be equal to the moment relative to a corresponding positive charge at the nitrogen nucleus of one of the electrons of the unshared pair occupying the fourth tetrahedral nitrogen orbital. See C. A. Coulnoa, Trans. Faraday SOC.,SO, 433 (1942), for n discuawion rhowing thnt effectr of thin kind caa be dgnificant.

AN ELECTRON-DIFFRACTION INVESTIGATION OF NITROCEXTRIFLUORIDE

March, 1950

I

I

I

I

0

20

40

Q

I

I

I

60

80

100

1183

+.

Fig. 1.-Simplified theoretical intensity curves for symmetrical pyramidal models of nitrogen trifluoride with the N-F distance 1.37 A. and the indicated bond angles. The arrows indicate the measured positions of features of the diffraction pattern.

scribed below, from the nature of the charge distributions used to explain the observed dipole moments. It may be noted that there is no obvious regularity which suggests a correlation between the observed bond angles and either the steric forces or the coulombic repulsionsD arising from the ionic character of the bonds which might be expected in these examples. Thus, the ionic character of the N-H bond is expected t o be about equal to that of the N-F bond, with changed sign, while i t seems probable that the steric interactions, if repulsive, are greater for F...F than for H...H. Neither is there a n obvious general correlation with the strengths of the bonds, such as would be significant because the strengths of the bonds in some degree determine the hybridization of the nitrogen or oxygen bondforming orbitals and, hence, the bond angle. If the distribution of charges invoked in the electric moment discussion is considered, however, i t is seen that in nitrogen trifluoride, for example, the repulsions among the negatively charged fluorine atoms are almost balanced by the repulsions between the fluorine atoms and a center of negative charge (representing the uncompensated member of the unshared pair of electrons) on the backside of the nitrogen atom, whereas in ammonia the repulsions among the positively charged hydrogen atoms are augmented by the attractions between the hydrogen atoms and the negative ( 0 ) Reference 6, p. 78.

backside of the nitrogen atom. Consequently, the bond angle value adopted by the nitrogen trifluoride molecule is only slightly increased by these coulombic interactions over that which it would have in their absence, while in ammonia there may be a considerable increase due to these forces. Quantitative calculations of this effect, carried out with reasonable assumptions regarding the magnitude and disposition of the charges involved and the force constant for symmetrical bending of the bond angles, gave results of the right magnitude ; according to these calculations i t is not improbable that the increase in the bond angle due to these coulombic forces is several degrees in ammonia, but only a fraction of a degree in nitrogen trifluoride. This interpretation of the bond angles and dipole moments of nitrogen trifluoride and ammonia must be applicable also to other series of pyramidal or bent molecules involving a given apical atom and a series of peripheral atoms of varying electronegativity. As has been mentioned] the usual bond-angle variations do agree with the interpretation; it is not so easy to see whether the dipole moment evidence will in general fit in as satisfactorily. The N-F bond distance in nitrogen trifluoride is 0.07 k. shorter than the distance 1.44 A. reportedlo for difluorodiazine, F-N-N-F, by Bauer. However, i t is just equal t o the value (io) s. E.BWW, amm JOUSHAL, w, aior (10471.

VERNER SCIIOMAKER AND CIIIA-SILu

1184

VOl. 72

given by the empirical equation

to a curve closely resembling the observed diffraction pattern. The values of qobs shown in the Table I are (IN =: 0.74 A., IF = averages, with an average deviation of 0.6% 1.0, (rN-F)oalcd. = 1.37 -k.) (excluding the Iirst ring), obtained from two sets which was found” by Schomaker and Stevenson of measurements of the ring diameters by each of to relate the length of a partially ionic single the authors. Except for the first, second, ninth, covalent bond to the normal covalent single bond tenth and eleventh rings, for which fewer readings radii of the component atoms and the difference were taken, each of these four sets included from of their electronegativities; it is appreciably five to twenty individual measurements made in greater than the sum, 1.34 Hi., of the Pauling- the usual way.’O The qobs values have been corrected for film expansion. Huggins radii for nitrogen and fluorine. Experimental Method and Results.-The TABLE I ,sample of nitrogen trifluoride used was provided Poaled./Fbs. 104O Weight” gob.. 1000 102 Max. Min. by Professor Anton B. Burg. According to Pro1 7.97 1.054 1.054 1.035 0 fessor Burg i t contained no significant amount of 1 12.39 1.037 1.017 0.977 0 any impurity. The sample was condensed t o a 2 ,945 15.77 0.977 0.951 0 liquid, an appreciable fraction was pumped off 2 ,963 20.76 0.978 0.973 0 through the apparatus before making the pictures, 3 25.95 1.004 0.996 ,983 3 and the pictures were made with gas in equilibrium 3 30.89 1.021 1.010 1.002 3 with the remaining liquid a t temperatures deter4 2 35.61 1.038 1.026 1.011 mined by the required pressure. Fifteen dif4 39.57 1.027 0.998 0.975 2 fraction pictures of various densities were made 5 43.94 0.999 0.987 .976 2 with the apparatus described by Brockway.12 48.75 1.007 1.003 .989 5 3 The electron wave length, 0.0610 A. on the Sieg6 ,999 53.84 1.020 1.009 3 bahn scale, was corrected for the (measured) ,999 6 59.15 1.019 1.009 2 expansion of the photographic film used in making 7 62.62 1.049 1.033 1.006 0 the zinc oxide and gold calibration pictures. 7 66.27 1.038 1.008 0.994 0 The pictures show as many as eleven rings with 8 .996 71.45 1.015 1.005 2 an apparent intensity distribution which is very ,997 8 76.72 1.015 1.006 3 well represented, except at large scattering angles, 9 .993 82.6 1.011 1.004 2 by the theoretical intensity curves “102’” and 87.0 1.024 1.014 1.000 1 9 “104’” of the Fig. 1. At large angles the appear10 90.2 1.044 1.018 0.991 1 ance of the sixth and seventh maxima and the 10 94.9 1.021 0.999 .989 1 observed equality of the ninth and tenth maxima 11 .997 99.6 1.016 2 1.006 would correspond perfectly to an interpolated 11 2 .991 105.7 1.011 1.001 curve for a bond angle of about 102.5’. The simplified theoretical intensity curves shown in All features Av. 1.019 1.006 0.991 the Fig. 1 are plots of the functionls Av. dev. 0.0142 0.0129 .0122 with equal

- 0.09 -xg/ 0.72 A., (XF- X x / =

*A-B = r A

f

rB

WtS.

Quoted wts. 10 - s,

is the scattering angle and the 2’s are the atomic members) corresponding to rigid, symmetrical, pyramidal models of the nitrogen trifluoride molecule with various bond angles. We conclude that to the precision afforded by our diffraction data (extending to p = 106) the nitrogen trifluoride molecule is such a symmetrical triangular pyramid. A model with the unreasonable parameters YN-F = 2.14 8.,rF-F = 1.37 8.,and L F N F = 37.3’ would agree well with the data because of the similarity in scattering power of the nitrogen and fluorine atoms, but i t is not possible that any other symmetrical model with a bond angle widely different from those represented in the Fig. 1 should lead (p = ?sin

‘P =

A

‘P

(11) V. Schomaker and D. P. Stevenson, THISJOURNAL, 68, 37 (1941). (12) L. 0.Brockway, Rev. Modern Phys., 6, 934 (1986). (18) R. Spurr and V. Schomaker, THIOJOURNAL, 64,2696 (l942), second pusgrsph, footnote 6,

Av. Av. dev. Av. Av.dev.

1,017 1.005 0.0076 0.0060 1.016 1.005 0.0079 0.0065

* 993 .0073 * 993 ,0089

Quoted wts. 3rd min. t o 8th max. only Five features Av. 1.013 1.005 ’ .994 .0064 with triple Av. dev. 0.0064 0.0041 wt., only Best value 1.016 1.005 0.993 1.392 1.377 1.360 rN- F 2.134 2.141 2.145 rN- F a These weights are based on the probable reliability of the measurements, rings close in or very far out and unsymmetrical or doubled rings being considered relatively unreliable. 1

The quantitative comparison of these values with the positions of the maxima and minima of the theoretical curves is given in the table for bond angles of 100, 102 and 104’ and is summarized by means of averages and average deviations calculated with various weightings of the p/pob, values. The best consistency is found for

SOMEHEATSOF WETTINGOF UNITSURFACES

March, 1950

102' with perhaps a slight tendency toward 104'. No significant uncertainty in the values of q/qohs is introduced by the question of the best weights to be used in taking the average. From the qualitative comparisons and these consistency measures we believe it is very unlikely that the correct bond angle value lies outside limits given by our final result L F N F = 102.5 * 1.5 . Our final values for the nitrogenfluorine distance I N - F = 1.37 =t0.02 8. and the fluorine-fluorine distance I F . . . F = 2.14 * 0.02 A. are taken from plots of the values of YN-F and Y F . . .F (from Table I) as functions of the assumed bond angle. The quoted uncertainties for these quantities we also regard as limits of error within which the correct values almost certainly lie. They include the variation corresponding t o the quoted uncertainty of the bond angle and the estimated uncert%intyof 4 / q o h s ( 0.007, corresponding to *0.01 A. in Y N - F ) for the given bond angle. Acknowledgment.-We are indebted to Professor Burg for the sample of nitrogen trifluoride f

[CONTRIBUTION FROM

THE

1185

and to Professor Pauling, especially for discussion of the dipole moment problem. The late Dr. Horace Russell, Jr., helped us in making the photographs. Summary

An electron-diffraction investigation of nitrogen trifluoride has been made. The molecule may be represented by a symmetrical, pyramidal model with bond distance I N - F = 1.37 A. and b p d angle L F N F = 102.5' ( I F . . . F = 2.14 A.), values which are almost certainly in error by less than *0.02 8. for the distances and +1.5' for the bond angle. A simple interpretation, which should be applicable t o other, analogous situations, is given for the very small dipole moment and relatively small bond angle of the nitrogen trifluoride molecule, in contrast to the relatively large dipole moment and bond angle of the ammonia molecule. PASADENA, CALIFORNIA

RECEIVED AUGUST 5, 1949

LABORATORY OF PHYSICAL CHEMISTRY, THESTATE COLLEGE O F

WASHINGTON]

Some Heats of Wetting of Unit Surfaced BY FRANK L. H O W A R D AND ~ J. L. CULBERTSON Introduction Previous studies in this Laboratory and elsewhere have shown interesting correlations between heats of wetting of various solid-liquid systems and such other properties as the apparent density of the pulverized solid,s polarity,* adsorptive ~ a p a c i t y ,coking ~ quality of coalI6 and hygroscopicity of soils.' Most of the available data have been reported in terms of the weight of solid wetted. However, in the study of surface phenomena it is desirable that the pertinent thermodynamic data be based on the surface area involved.8 Harkins and c o - w o r k e r ~ have ~ ~ ~ ~reported ~~ heats of emersion (the reverse of heats of wetting, (1) From a thesis submitted by Frank L. Howard to the Graduate School of The State College of Washington in partial fulfillment of the requirements for the degree of Doctor of Philosophy, June, 1948. This paper was presented at the San Francisco Meeting of the American Chemical Society, April, 1949. (2) Present address: Physical Chemistry Department, Division of Metallurgical Research, Kaiser Aluminum & Chemical Carp., Spokane, Wash. (3) Culbertson and Winter, THIS JOURNAL, 59, 308 (1937). (4) Boyd and Harkins, i b i d . , 64, 1190 (1942). ( 5 ) Harman and Parmelee, J . A m . C c r a n . Soc., 28, 110 (1945). (6) Cannon, Griffith and Hirst, Proc. Conf. Ultra-fine Structure of Coals and Cokes, Brit. Coal Utilization Research Assoc., 1944, 131; Chrm. Abstracts, 89, 1274.2 (7) Behrens, Z. PAaxrcncrnarh., XWngurg Y. Bodcnk., 40, 267 (1935). (8) Harkins and Boyd, THIDJOWSNAL, 64, 1195 (1942). (0) Basford, Jura and Harkins,ibid., TO, 1444 (1948). (10) Jura and Harkins, ibid., 66, 1866 (1044).

as used here) on this basis for a number of systems, and have discussed in detail the theory and the experimental difficulties of such measurements. This investigation was designed to measure the heats (enthalpies) of wetting of solid surfaces in terms of the wetted surface area. Experimental The solids chosen were the sulfides galena and sphalerite, the halides sylvite and fluorite, and quartz. With one exception, the solids employed were selected mineral crystals; these were reduced by dry milling t o powders with specific surface area ranging, approximately, from 0.1 t o 2 square meters per gram. (The sylvite was C. P. KC1.) Water, n-heptane and carbon tetrachloride as wetting liquids represented polar, non-polar and polarizable types, respectively. The pulverized solid sample, of from 5 t o 20 g., was weighed into the powder chamber of the wetting apparatus (Fig. l),evacuated a t 300" for an hour, then sealed under vacuum. The liquid reservoir chamber was filled with the wetting liquid, plus enough powder to ensure saturation with the solid phase. This precaution was deemed desirable to avoid any possible error from heat of solution of the solid phase in the liquid. Carbon tetrachloride and n-heptane, when used, had activated alumina or sodium amalgam added t o maintain their anhydrous condition. A loop of fine constantan wire was engaged with the hookshaped septum between the reaction and reservoir chambers. The wetting apparatus was then inserted in the calorimeter and permitted to come to thermal equilibrium a t 25". The adiabatic calorimeter comprised two concentric metal vessels, the inner one containing the wetting apparatus, while the outer one carried a compensating heating coil which served to maintain the temperature equality